[R-lang] Help with screening data before glmer

[Francesco]

I'm trying to find a suitable way of plotting the relationship
between 2 fixed effect categorical IVs (Syntax and Semantics),
2 random effect factors (ID and item) and a
categorical DV (Correct). Syntax and Semantics have two levels
each.

The point is I would like to look
at the distribution of the response (log-odds in this case),
considering one fixed effect at a time, plotted for each participant
or each item (in xyplot-type style). For example, I could look
at the log-odds of correct responses by syntax for each participant
or for each item, reason being I would like to look at potential
skewness of data, outliers, awkward responses and not
look at model fit (i.e. residuals).

There seems to be quite a bit available if you have a continuous
predictor http://www.ats.ucla.edu/stat/r/dae/melogit.htm
but nothing for discrete ones.

When plots based on frequency tables are brought in, they do not
allow grouping variables (random factors) like ID and item.

Any help is much appreciated.

-- 
Frank Romano

Sabanci University
website: http://sabanciuniv.academia.edu/FrancescoRomano